Background Variables, Levels of Aggregation, and Standardized Test Scores

Abstract

This article examines the role of student demographic characteristics in standardized achievement test scores at both the individual level and aggregated at the state, district, school levels. For several data sets, the majority of the variance among states, districts, and schools was related to demographic characteristics. Where these background variables outside of the control of schools significantly affected averaged scores, and test scores result in high stakes consequences, benefits and sanctions may be inappropriately applied. Furthermore, disaggregating the data by race, SES, limited English, or other groupings ignores the significant confounding and cumulative effects of belonging to more than one disadvantaged group. With these approaches to evaluation being fundamental to the No Child Left Behind mandates, the danger of misinterpretation and inappropriate application of sanctions is substantial.